GitHub announced Agent HQ, a new centralized platform that integrates multiple AI coding agents. The initiative, revealed yesterday at the GitHub Universe 2025 event, will allow developers with a paid subscription to GitHub Copilot to access and manage tools from companies such as Anthropic, OpenAI, Google, Cognition and xAI, in addition to Copilot itself.
The goal is to unify the development workflow, currently divided by different AI interfaces. The new feature will be managed by a control panel (mission control), where developers will be able to assign tasks, guide and track the work of multiple AI agents.
What does GitHub’s new AI hub look like?
The idea is to offer an ecosystem that brings together different agents in one place. The platform will allow developers to orchestrate “fleets of specialized agents” to execute complex tasks in parallel. It will be possible to track the work of multiple AI agents simultaneously, including running several of them in parallel on the same task to compare results.
The system also includes new branch controls to oversee when to run continuous integration (CI) checks on AI-generated code, as well as identity capabilities to manage access and policies for each agent, treating them as if they were a human developer working on a team.
Mission Control will also connect to project management tools Slack, Linear, Atlassian Jira, Microsoft Teams, and Azure Boards.
When will agents be available?
GitHub reported that Copilot Pro+ users participating in the VS Code Insiders program will be able to immediately access OpenAI Codex, making it the first agent available. The rest will be made available in the coming months, as part of the paid subscription.
What else was announced?
Along with Agent HQ, GitHub introduced a set of tools focused on planning, customization, and enterprise governance. In VS Code, “Plan Mode” was introduced, which asks the developer questions to help build a detailed approach before starting code production. After the plan is approved, it is sent to Copilot or another agent for implementation.
For enterprises, a public preview of GitHub Code Quality has been released, which analyzes code maintainability, reliability, and test coverage, and metrics dashboarda dashboard for administrators to track the impact and usage of Copilot in the organization.
A code review step has also been added to the Copilot agent workflow, allowing it to access tools, such as CodeQL, to evaluate code before passing it to a human developer. Finally, a governance “control plane” will allow corporate administrators to define security policies, record audits, and manage which AI agents are and are not allowed within a company.
Source: https://tecnoblog.net/noticias/github-libera-o-acesso-a-varios-agentes-de-ia-como-claude-e-codex/
